Screening for double carbapenem therapies effective against multidrug resistant Enterobacteriaceae

(1)

Screening for double carbapenem

therapies effective against multidrug

resistant Enterobacteriaceae

Karin Vickberg

____________________________________________

Master Degree Project in Infection Biology, 45 credits. Spring 2019

Department: Medical Biochemistry and Microbiology

(2)

(3)

Abstract

Multidrug resistant bacteria account for one of the biggest threats to human health in the future. One treatment option to be evaluated is antibiotic combination therapy. For this purpose, last-line carbapenems have been suggested, even when bacteria are resistant, e.g. via production of degrading enzymes called carbapenemases. This study aimed to analyze combinations of ertapenem, doripenem and meropenem at clinically relevant concentrations against strains of Escherichia coli and Klebsiella pneumoniae carrying at least one carbapenemase (NDM, KPC or OXA-48). The screening methods used were a relatively new time-lapse microscopy method (oCelloScope), spot tests and time-kill experiments. The oCelloScope uses algorithms to illustrate bacterial growth over time. Moreover, a spot test involves dilution and plating of oCelloScope samples at 24 h. In the time-kill assay, samples are plated from tube cultures at specific time points. Depending on the difference in growth between antibiotic combinations and the most potent monotherapy, the effect was classified as synergistic, antagonistic or indifferent. According to the spot test, synergistic effects were observed against 12/30 strains (mainly carrying OXA-48) with one or more antibiotic combinations. A possible cause for this pattern is the lower activity of OXA-48, leading to lower resistance levels. Synergies also existed against some of the highly resistant strains (carrying NDM or KPC), which can be of great importance for clinical treatment. Due to a higher resolution of the spot test, the oCelloScope was unable to detect some of the synergistic combinations that were detected with the spot test. For future studies, more strains should be tested with time-kill, since only three strains were tested without observing any synergistic effects.

Key words and abbreviations

Combination therapy Carbapenem resistance Carbapenemase

Carbapenem-resistant Enterobacteriaceae (CRE) Carbapenemase-producing Enterobacteriaceae Klebsiella pneumoniae carbapenemase (KPC) New Delhi metallo-b-lactamase (NDM) Oxacillinase (OXA)

Ertapenem (ETP) Doripenem (DOR) Meropenem (MEM)

Broth microdilution (BMD)

Time-lapse microscopy (oCelloScope) Time-kill experiments

(4)

Popular summary

Who could have thought that bacteria (of all things) will account for one of the biggest threats to human health in the future? Bacteria have become resistant to many of the antibiotics we use today, and it is not uncommon that the same bacterium can resist several different antibiotics (multidrug resistant bacteria). It is scientifically difficult to discover new antibiotics, and the process is both long and expensive. If we are not dealing with the problem with multidrug resistant bacteria, common infections might be impossible to treat in the future.

In order to treat infections caused by multidrug resistant bacteria, old antibiotics that have been discarded as treatment options due to toxicity problems and/or low efficacy have now been re-introduced. The infections have also been treated by using regimens consisting of a combination of two or more antibiotics. Even if the bacteria are resistant to the antibiotics when administered alone, they sometimes cannot withstand both antibiotics at the same time (if this happens, the antibiotics are said to have a synergistic effect). Some antibiotics can enhance bacterial growth only if they are given together, which is called antagonism. The research area with antibiotic combination therapy is relatively new, and different combinations have to be tested against multidrug resistant bacterial strains.

In this project, combination therapy with the antibiotics ertapenem, doripenem and meropenem was tested against two different bacterial species; Escherichia coli and Klebsiella pneumoniae. Both these bacterial species contained molecules that were able to break down the antibiotics (i.e. carbapenemases). During the last years, a new method has made it a lot easier to study antibiotic combinations. This method includes an instrument called the oCelloScope. In this study, the oCelloScope was used in combination with so-called spot tests and time-kill assays in order to screen for synergistic effects between antibiotics. The oCelloScope contains a camera that repeatedly takes images of the sample, which then can be used to determine bacterial growth. A low growth means that the antibiotics have killed the bacteria. After the oCelloScope was finished, a spot test was performed, which means that the bacterial growth was checked on plates containing nutrients for the bacteria. By using spot test, the resolution is increased. Lastly, the time-kill assay includes cultures that contain different concentrations of antibiotics and combinations, wherefrom samples were taken at specific time-points in order to determine the change of growth over time.

(5)

1 Introduction and background

Antibiotic resistance has become a big problem in the world. It is not only causing a major economic burden, but also an increased morbidity and mortality.1_{According to Cassini et al.}

(2018), antibiotic resistant bacteria might even account for one of the biggest threats to human health in the future.2_{The incidence of infections caused by antibiotic resistant bacteria is}

increasing, and there are microbes present (multidrug resistant (MDR) bacteria), that have developed resistance to almost every available antibiotic.3_{The reason why resistance has arisen}

is e.g. an unnecessary and inappropriate use of antibiotics.4_{Antibiotics have not only been used}

as treatments, but also within food production in order to promote animal growth. Using antibiotics unnecessarily and in farms end up with antibiotics in soil and groundwater – an environment suitable for constant selection of MDR bacteria.5

When new antibiotics enter the market, resistance is often developed shortly after the distri-bution.1_{As it is today, it is scientifically difficult to discover new antibiotics. A big part of the}

industry is no longer investing in finding new classes of antibiotics, since the market is relatively unprofitable.6_{Hence, it is important that existing antibiotics are used as little as}

possible. This can be done e.g. by implementing rapid diagnostic tests in order to decrease the unnecessary use of drugs. Another important factor is to reduce the transmission of infections, so that antibiotics are less needed. Infections can be contained by using precautions and having well-structured government agencies to whom notifiable diseases can be reported.1,7_Except

from using already existing antibiotics to a lesser extent, new antibiotics are formed by modi-fying old ones in order to outrun the resistance. However, resistance is easily developed and new classes of antibiotics would be preferred.1_{As it looks today, new alternative treatments are}

needed to treat infections caused by MDR bacteria.

1.1 Alternative treatments against MDR bacterial infections

To date, a lot of research is ongoing regarding alternative treatments against infections caused by MDR bacteria. Therapies based on antibiotic combinations have been suggested, since bacteria in some cases can show resistance to antibiotics when used alone, but show susceptibility when the antibiotics are combined. In that way, the bacteria can potentially be eliminated. Furthermore, other alternative treatments that have been suggested are bacteriophages (or simply phages) or inhibition of virulence factors. Phages can kill bacteria with high specificity and by using cocktails of phages, the risk for resistance development is reduced.1,4_{Until 2016, no phages were used in a therapeutic practice.}1_{However, in a study by}

Scooley et al. (2017), a patient suffering from necrotizing pancreatitis was treated with phage cocktails in the US. The patient became progressively worsened despite antibiotic treatment, and he survived thanks to the phage therapy. After eight weeks, the patient was considered as healthy.3_{Regarding inhibition of virulence factors as an alternative treatment to antibiotics, a}

study was performed by Heidari et al. (2017) on Pseudomonas aeruginosa. The authors used the chemical compound pyridoxal lactohydrazone with the aim of inhibiting the quorum sensing (QS) system of the bacterial cell. A bacterium uses its QS system in order to communicate with other cells, and the QS system can hence be used in order to induce expression of virulence factors. If the QS system is inhibited, so is the expression of virulence factors.8_{However, both}

(7)

1.2 Resistance to last-line treatments

Of all present antibiotic classes, carbapenems are used as a last-resort for treating infections caused by otherwise drug-resistant bacteria. Carbapenems belong to the broad-spectrum antibiotic class b-lactams, which is the most commercially available antibiotic class on the market.10,11_{The b-lactams are both effective and safe to use, and the common structure among}

them is the b-lactam ring. However, resistance has arisen also to carbapenems. There are different resistance mechanisms – the most widespread one is via the production of enzymes called carbapenemases.10,12_{These enzymes are b-lactamases that are capable of hydrolyzing}

carbapenems by attacking the b-lactam ring.11_{The most frequently encountered}

carbapenem-ases that are most clinically relevant (i.e. KPC, NDM and OXA-48) can be seen in Table 1.7,12,13

Interestingly, some OXA enzymes have carbapenem-hydrolyzing activity (e.g. OXA-48) and some do not (e.g. OXA-1 or OXA-9).25,26_{Furthermore, other carbapenem resistance}

mecha-nisms are porin deficiencies or production of efflux pumps. Porin deficiencies (e.g. of OmpC and OmpF in Escherichia coli) lead to a decreased entry of b-lactam antibiotics into the bacterial cell. The presence and an enhanced production of efflux pumps lead to an increased outflow of antibiotics from the cell. Additionally, bacteria that have several resistance mechanisms are more prone to be highly resistant against carbapenems.10,13

Table 1

Summary of the most frequently encountered carbapenemases (KPC, NDM and OXA-48) in

Enterobacteriaceae and which antibiotics they can hydrolyze.7,12,13

Enzyme Abbre-_viation Host organisms

Enzyme substrates

Penicillin Cephalo-_sporins Aztreonam _penems

Carba-Klebsiella pneumoniae

carbapenemase KPC

K. pneumoniae, E. coli + other

Enterobacteriaceae Yes Yes Yes Yes

New Delhi

metallo-b-lactamase

NDM Enterobacteriaceae Yes Yes No Yes

Oxacillinase OXA-48

K. pneumoniae +

other

Enterobacteriaceae Yes No* No Yes

*The organism carrying the OXA-48 gene often also carries an extended-spectrum-b-lactamase (ESBL), making them resistant to cephalosporins as well.12

Resistance to most clinically available carbapenems can be seen with many Gram-negative bacteria, such as E. coli and Klebsiella pneumoniae – bacteria belonging to the family Enterobacteriaceae.10_{In fact, carbapenem-resistant Enterobacteriaceae (CRE) are seen as one}

of the biggest threats to human health – especially E. coli and K. pneumoniae, which are the most frequently encountered carbapenem-resistant species via carbapenemase production.7,12

There is an increasing spread of carbapenemase-producing bacteria globally, which is of high concern because of the higher patient morbidity and mortality.7_{Figure 1 shows the global}

spread of carbapenemase-producing K. pneumoniae in 2016, and as seen in the figure, these bacteria can be found in most parts of the world.14_{Furthermore, the risk factors for getting}

infected with CRE involve e.g. hospitalization, using catheters, transplantation or an epidemiological connection to a patient colonized by CRE.15,16_{CRE can lead to an}

(8)

Figure 1

The global spread of carbapenemase-producing K. pneumoniae in 2016. The carbapenemase genes included were blaKPC, blaNDM and blaOXA-48-like. Colored countries correspond to a record of carbapenemase(-s), which

includes endemic or sporadic spreads. Gray colored countries mean that neither KPC, NDM nor OXA-48-like have been recorded. The figure is re-constructed from Lee et al. (2016).14

If resistance is observed against carbapenems, there are often few treatment options. As seen in Table 1, some carbapenemases (i.e. KPC and NDM) can also hydrolyze the b-lactams penicillin and cephalosporins.12_{Actually, most of the carbapenemase-producing Enterobacteriaceae are}

resistant to all antibiotics in the b-lactam class and to many other antibiotics as well, due to additional b-lactamases carried on the same plasmid.18,19_{In order to treat the infections,}

anti-biotics that have been rarely used historically due to efficacy and/or toxicity problems (e.g. polymyxins, tigecycline and fosfomycin) have been re-introduced.11_MDR

carbapenemase-producing K. pneumoniae can often only be treated with polymyxins, tigecycline or fosfomycin, but resistance is emerging also to these agents.20_{It goes without saying that infections caused}

by MDR Enterobacteriaceae are difficult to treat and are associated with serious treatment failures. In turn, treatment failures lead to delayed time periods until a successful therapy is achieved – and the mortality rate increases.21

1.3 Antibiotic combination therapy

As resistance has been developed both against carbapenems and the old antibiotics that have been re-introduced, there is an urgent need for new treatment options. Many researchers have been tempted by the idea of antibiotic combination therapy, so the research area has started to expand. However, available data are still limited as few studies have been performed and published. Interestingly, in a study by Bulik & Nicolau (2011), a combination of two carba-penems (ertapenem and doripenem) was tested. The combination showed a decrease in bacterial density compared to monotherapy – both in vivo and in vitro.22_{A successful outcome was also}

(9)

meropenem, showed less bacterial growth than monotherapy against 3/8 strains of K. pneumoniae carrying KPC, NDM or OXA-48.19_{Furthermore, another example of a previous}

study is presented by Fredborg et al. (2017), where a synergistic effect (> 2 log2 decrease in

minimum inhibitory concentration compared to the most effective monotherapy) was seen against 4/5 strains when combining ertapenem and meropenem.21_{In other words, further}

research regarding carbapenem combinations is warranted and more data need to be published. The mechanism of action of carbapenems is to acylate penicillin-binding-proteins (PBPs), making them inactive (Figure 2). PBPs are involved in the formation of peptidoglycan – a compartment of the cell wall. Due to the inactivation of PBPs, the cell wall gets weakened and the cell will eventually burst due to osmotic pressure. There are different kinds of PBPs present in a bacterium, and carbapenems can often bind to several of them.10_{One theory behind}

carbapenem combination therapy is that ertapenem serves as a decoy for the other carbapenem by binding to the carbapenemase because of a high binding affinity.12_{Doripenem and}

mero-penem are more potent because of a less susceptibility to hydrolysis by carbamero-penemases. Since ertapenem consumes most of the carbapenemases and also blocks the binding sites, doripenem or meropenem can bind to the PBPs and eventually lead to cell death.10,20,22

Figure 2

The theoretical mechanism of using carbapenem combinations in order to treat infections caused by MDR

Enterobacteriaceae. Ertapenem and either doripenem or meropenem enter the bacterium through

outer-membrane porins. In the periplasmic space, ertapenem binds to the carbapenemases, due to a high binding affinity. Because of this binding, the degradation of the other carbapenem is prevented. Doripenem or meropenem can instead bind to the penicillin-binding-proteins (PBPs), which normally are involved in the formation of peptidoglycan layers of the bacterial cell wall. Because of the carbapenem binding, PBPs are no longer capable of contributing to the cell wall formation, and the bacterium dies. The figure is re-drawn from Blair et al. (2015) by Karin Vickberg © 2019.23

1.4 Time-lapse microscopy (oCelloScope)

(10)

standard method for these kind of screening purposes is time-kill experiments, but that is very time-consuming and labor-intensive. Meanwhile, in the oCelloScope, 96 samples can be tested simultaneously (Figure 3A). The instrumental principle is to determine the amount of bacterial growth in the wells by generating series of images through microscopy scanning. The system is built up of an illumination unit, a lens and a camera (Figure 3B). Images taken with the camera can be processed into two-dimensional (2D) pictures (Figure 3C) and translated into bacterial growth curves by using growth kinetics analysis (GKA) algorithms. Two GKA algo-rithms are the background corrected absorption (BCA) and Segmentation and Extraction of Surface Area (SESA). BCA is a measurement of the intensity of dark objects (bacteria), taking the background intensity into consideration. A threshold value is used to separate image pixels into dark object pixels and background pixels. In other words, the BCA value would increase with a higher bacterial concentration, as the intensity of dark objects increases. SESA on the other hand, is not affected by differences in light intensity. It uses segmentation and identifies bacteria based on their contrast against the background. As bacteria start to overlap, SESA will be less reliable at higher concentrations.31–33

Figure 3

The oCelloScope device. A) A picture of the oCelloScope device with a 96-well microplate inserted. B) An overview of the principle for bacterial detection. An illumination unit puts light onto each well while the camera takes images from underneath with an angle of 6.25° against the horizontal plane. C) The images can be processed into two-dimensional pictures, here showing the bacterium Staphylococcus alactolyticus. The figure is used with permission from Fredborg et al. (2013).31

1.5 Project aim

In this project, different combinations of the carbapenems ertapenem, doripenem and mero-penem were tested in vitro against multidrug resistant Enterobacteriaceae. In this case, clinical E. coli and K. pneumoniae were used, carrying at least one carbapenemase (KPC, NDM and/or OXA-48). An oCelloScope was used to screen for synergistic effects, and so-called spot tests and time-kill assays were used to further evaluate the results. The objective was to contribute to the research regarding carbapenem combinations at clinically relevant concentrations, by using a total of 30 strains with different sets of enzymes. Furthermore, the main goal was that this study can improve the situation with MDR Enterobacteriaceae by adding more knowledge, as combinations of carbapenems potentially can be used to treat serious infections caused by these highly resistant bacteria.

B

A

B

(11)

2 Materials and methods

2.1 Media and bacterial isolates

For all experiments in this project, cation-adjusted Mueller-Hinton (MH-II) broth and agar plates were used. All bacterial strains were sent from the Public Health Agency of Sweden, and the strains were first isolated at Swedish hospitals from clinical cases. When all samples were received, they were analyzed with the rapid laboratory method MALDI-TOF mass spectrometry (Bruker MALDI Biotyper). A total of 30 strains were included in this project – 14 E. coli strains and 16 K. pneumoniae strains. Each strain was given a specific Antibiotic Research Unit (ARU) number. All strains were screened in the time-lapse microscope (oCelloScope) and with a spot test, and three strains (ARU724, ARU923 and ARU928) were also analyzed by time-kill experi-ments. Moreover, four strains (ARU735, ARU882, ARU891 and ARU894) were included in a carry-over test (section 2.8), which determines if the lack of growth is due to actual elimination of bacteria or due to a lasting effect of antibiotics. The reference strain used for positive controls in all experiments was P. aeruginosa (ATCC®₂₇₈₅₃TM_{), recommended by EUCAST.}24

The number of E. coli carrying either KPC, OXA-48 or NDM was three, five and six, respect-ively. Moreover, the number of K. pneumoniae strains carrying either KPC, OXA-48 or NDM was five, three and five, respectively. Also, three K. pneumoniae strains carrying both NDM and OXA-48 were included in the project. Most strains also carried other b-lactamases besides the carbapenemase(-s), more specifically CTX-M, TEM, OXA, SHV, LEN and/or OKP.

2.2 Antibiotics

The antibiotics used in this study were ertapenem (ETP), doripenem (DOR) and meropenem (MEM). All antibiotics were purchased from Merck KGaA (Darmstadt, Germany). Preparation of stock solutions was performed by dissolving ertapenem in phosphate buffer (pH 7.2; 0.01 mol/L), doripenem in physiological saline (0.85%) and meropenem in sterile water to a stock concentration of 1280 mg/L. Stock solutions of antibiotics were stored at -80°C for maximum six months after preparation.27

(12)

32 mg/L for doripenem and 64 mg/L for meropenem. Moreover, for the antimicrobial suscepti-bility testing, the antibiotic concentration range reached from 0.031-32 mg/L, with steps of 1:2 dilutions in between.

2.3 Antimicrobial susceptibility testing using broth microdilution (BMD)

The minimum inhibitory concentrations (MICs) of the strains were determined with broth microdilutions (BMDs) recommended by EUCAST.29_{Microdilutions were performed with the}

antibiotics ertapenem, doripenem or meropenem in triplicates. Antibiotics were diluted in MH-II broth and loaded onto a round-bottomed 96-well microplate (Greiner Bio-One GmbH, Monroe, USA). Thereafter, a suspension of 3-5 distinct colonies was prepared by selecting colonies from a fresh agar plate. The colonies were suspended in sterile saline, and the suspension was adjusted to 0.4-0.6 McFarland turbidity standard by using a photometric device (DensiChek Plus, Biomérieux). Separate suspensions were used for different replicates run on the same day. Within 15 minutes of preparation, the suspension was diluted 1:100 times in MH-II broth. Each well in the microplate was inoculated with bacterial suspension (1:2 dilution) within 15 minutes after dilution. The microplate was incubated with a lid at 35 ± 2°C for 16-20 h. Meanwhile, the inoculum suspension was diluted 1:1000 times before plating for colony count, and the agar plate was incubated at 37°C overnight. After incubation, colonies were counted manually using a colony counter (Scan® 100 interscience, St Nom, France). The MIC value was read from the microplate and compared to the EUCAST clinical MIC breakpoints. The highest tested MIC value was 32 mg/L, meaning that a strain growing also at 32 mg/L was noted to have a MIC value >32 mg/L. The following clinical MIC breakpoints were used for Enterobacteriaceae: susceptible against doripenem if MIC ≤ 1 mg/L and resistant if MIC > 2 mg/L.28_{Susceptible against ertapenem if MIC ≤ 0.5 mg/L and resistant if MIC}

> 0.5 mg/L. Additionally, susceptible against meropenem if MIC ≤ 2 mg/L and resistant if MIC > 8 mg/L (intermediate in between susceptible and resistant).29

2.4 Whole genome sequencing (WGS)

Whole genome sequencing (WGS) was performed by the Public Health Agency of Sweden on IonTorrent in order to detect mainly b-lactamase genes. The sequences were assembled in CLC Genomics Workbench and submitted to ResFinder.30

2.5 Time-lapse microscopy (oCelloScope screening)

Firstly, the strain was taken from a -80°C frozen stock and streaked on an agar plate. After incu-bation overnight at 37°C, 2-4 colonies were inoculated in MH-II broth and incubated at 37°C shaking at 190 rounds per minute (rpm) for 15-18 h. The next day, the overnight culture was diluted 100-fold in MH-II broth and incubated at 37°C shaking at 190 rpm for 1.5 h. The culture was diluted 20-fold in order to get a starting inoculum of approximately 106_{colony forming}

units per mL (CFU/mL). To ensure suitable bacterial concentrations, samples for viable counts were taken before each experiment.

(13)

growth controls, 90 µL to single antibiotic wells and 80 µL to antibiotic combination wells. Thereafter, 10 µL of antibiotic was added to single antibiotic wells (Figure 4). Antibiotic combination wells were loaded with 10 µL of each antibiotic stock solution. The microplate was pre-warmed at 37°C for about 15 minutes before bacterial culture was inoculated into each well except negative controls, so that a total volume of 200 µL was reached. Additionally, a trans-parent quantitative PCR film (Sarstadt AG & Co., Nümbrecht, Germany) was used to cover the microplate before it was inserted in the oCelloScope. The oCelloScope was in turn placed inside a Thermo Scientific Heratherm Incubator (Thermo Fisher Scientific Inc., Göteborg, Sweden) kept at 37°C (static incubation).

Figure 4

Microplate layout of oCelloScope screening, in this case with antibiotics doripenem (DOR) and meropenem (MEM). Five strains could be tested simultaneously, as well as a reference strain and a negative control. The figure is illustrated by Karin Vickberg © 2019.

In the oCelloScope, the camera used the autofocus function in order to find focus approximately in the middle of each well close to the bottom. The running time was set to 24 h, during which five images were taken every 15 minutes for each well. The image distance was set to 4.9 µm and the illumination level was 150. The detection level in the oCelloScope was 105_{CFU/mL =}

5 log10 CFU/mL.32 Screening in the oCelloScope was performed in one replicate.

2.6 Spot test

At the end of the oCelloScope screening, a spot test was performed by diluting the samples in the oCelloScope microplate from undiluted to 10-7_{(Figure 5). The spot test was done by taking}

(14)

Figure 5

Overview of the spot test principle. The samples in the oCelloScope microplate were diluted and plated as spots. On the following day, it was noted at which dilution growth was seen, the amount of growth and any possible carry-over effect (i.e. mainly growth around the edges of the spot). The figure is illustrated by Karin Vickberg © 2019.

2.7 Software

BCA and SESAmax values were calculated in UniExplorer version 8.1.0.7424 (Philips BioCell

A/S, Allerød, Denmark). A custom script in Python was used to extract 24 h values of BCA and the SESAmax values (at any time point), as well as images and plots.

2.8 Evaluation of carry-over effects

Carry-over tests were performed on four strains (section 2.1), which means that it was determined whether the lack of growth on plates were due to an actual elimination of bacteria or due to a lasting effect of carbapenems. Firstly, an oCelloScope screening (section 2.5) was completed with all three combinations of ertapenem, doripenem and meropenem. After 24 h, the microplate-content was diluted up to 100-fold followed by a regular spot test (section 2.6). In parallel with the spot test, 50 µL from each well was also transferred to a MH-II agar plate for colony counts. The samples were left to sink into the plates for about 1 minute before being spread with glass beads. All plates were incubated at 37°C overnight. After incubation, the log10

CFU/mL values of the spot test were compared with those of the glass bead plating. The plates were also examined for inhibition zones.

2.9 Time-kill experiments

(15)

All tubes were incubated at 37°C shaking at 190 rpm. The samples were serially diluted in PBS down to maximum 10-7_,_{and 100 µL of appropriate dilutions (depending on the cloudiness of}

the content) was plated onto agar plates and spread with glass beads. If the content was transparent, 250 µL of sample was taken in order to plate the undiluted sample as well. Incubation of plates took place at 37°C for about 24 h, followed by manual recording of visible colonies. All experiments were performed in duplicates. The detection level for the time-kill assay was 101_{CFU/mL = 1 log}₁₀_CFU/mL.

Figure 6

Schematic workflow of the time-kill assay. A total of seven tubes were used with different concentrations of ertapenem (ETP), doripenem (DOR) and meropenem (MEM). Samples were taken at 0, 2, 6 and 24 h and plated at appropriate dilutions. After incubation overnight, the results were read in terms of colony forming units (CFU/mL). The figure is illustrated by Karin Vickberg © 2019.

3 Results

3.1 MALDI-TOF results, MIC values and b-lactamase genes

Before starting the actual screening for synergistic effects from antibiotic combination therapy, the strains were tested with MALDI-TOF, BMD and WGS. These tests were performed in order to be sure that the samples contained the expected species of bacteria, to determine the resistance levels to carbapenems and to detect which b-lactam genes the strains carried. BMD was performed according to EUCAST recommendations and the WGS was performed on IonTorrent by the Public Health Agency of Sweden. The results revealed that all samples con-tained the expected bacterial species. Furthermore, in Table 2, all strains used in this project are listed, as well as their carbapenemase(-s), any other b-lactamases, their MIC values against ertapenem, doripenem and meropenem as well as classifications according to EUCAST clinical breakpoints.28,29_{It was shown that all strains were resistant to ertapenem according to the}

(16)

The b-lactamase genes were analyzed by assembling the sequences in CLC Genomic Work-bench and submitting them to ResFinder, in order to detect deviations from the reference sequences, e.g. mutations or insertions. Two strains were shown to carry a carbapenemase that deviated from the reference sequences. A more common event was the carriage of other b-lactamase genes with deviations from the reference sequences, which could be seen with 15/25 strains – mainly the SHV b-lactamase. Five strains did not carry any other b-lactamase genes besides a carbapenemase. The most common gene to carry was CTX-M (16/25 strains), followed by SHV (15/25 strains) and TEM (14/25 strains). Moreover, the strains carried in average 2-3 unique b-lactamase genes besides the carbapenemase gene.

3.2 Analysis of oCelloScope data and spot test

In order to analyze the effect from treating with carbapenem combinations in vitro, the algo-rithms BCA and SESAmax were used to illustrate bacterial growth in the oCelloScope

screenings. Because of the less reliability of SESA at higher bacterial concentrations, the maximum SESA value at any time-point (SESAmax) was used in the analysis of this project.

The BCA cut-off value at 24 h was ³ 8.00 and the SESAmax cut-off value was ³ 5.80, based on

a previous study.34_{Only if both BCA and SESA}

max were fulfilled, the result was defined as

growth. During growth inhibition, the BCA value remains stable or increases slowly over time, since dead cells also can be detected. Additionally, if two single antibiotic wells showed growth while the antibiotic combination at the same concentrations did not show any growth, the effect from the combination was defined as synergistic. Antagonism between the antibiotics was defined in case of growth of the combination, but no growth with one or both single antibiotics at the same concentrations. Finally, indifference was defined if growth was seen both for the antibiotic combination well and for the monotherapies at the same concentrations.

Regarding the spot test colony counts, the classifications synergy, antagonism and indifference were defined based on the difference in log10 CFU/mL values from the combination compared

to single antibiotic wells. Synergy was defined if the combination showed a ³ 2 log10 decrease

in CFU/mL at 24 h, compared to the most potent single antibiotic at the same concentration. The definition of antagonism was a ³ 2 log10 increase in CFU/mL, and indifference as a

< 2 log10 decrease or increase in CFU/mL at 24 h. In addition, a bactericidal effect was defined

as a ³ 3 log10 decrease in CFU/mL after 24 h of the combination compared with the start

(17)

Table 2

Summary of all strains, their carbapenemase(-s) and other b-lactamase genes, as well as their minimum inhibitory concentrations (MICs) against ertapenem (ETP), doripenem (DOR) and meropenem (MEM). MIC values were compared to clinical MIC breakpoints for susceptibility (S), intermediate (I) and resistant (R) according to EUCAST.28,29

Strain Species Carbapenemase _gene(-s) Other b-lactamase _genes MIC (mg/L)

ETP DOR MEM

ARU887 E. coli KPC-2 - 2 (R) 0.25 (S) 0.25 (S)

ARU888 E. coli KPC-2 - 8 (R) 1 (S) 2 (S)

ARU894 E. coli KPC-2 TEM-1A*, OXA-9* 16 (R) 1 (S) 2 (S)

ARU711 E. coli NDM-1 CTX-M-27 32 (R) 32 (R) 32 (R)

ARU713 E. coli NDM-1 CTX-M-27 >32 (R) >32 (R) 32 (R)

ARU714 E. coli NDM-1 CTX-M-27 >32 (R) 32 (R) 32 (R)

ARU892 E. coli NDM-1 - 32 (R) 16 (R) 32 (R)

ARU709 E. coli NDM-5* CTX-M-15, TEM-1B, _{CMY-2, OXA-1} >32 (R) 32 (R) 32 (R)

ARU717 E. coli NDM-5 TEM-1B >32 (R) 32 (R) 32 (R)

ARU889 E. coli OXA-48 - 4 (R) 1 (S) 1 (S)

ARU890 E. coli OXA-48 - 4 (R) 1 (S) 1 (S)

ARU891 E. coli OXA-48 TEM-1B 2 (R) 1 (S) 0.5 (S)

ARU896 E. coli OXA-48 CTX-M-15 4 (R) 0.25 (S) 0.5 (S)

ARU903 E. coli OXA-48 CTX-M-15, TEM-1B, _OXA-1 2 (R) 0.25 (S) 0.5 (S) ARU737 K. pneumoniae KPC-2 SHV-40* >32 (R) >32 (R) >32 (R) ARU869 K. pneumoniae KPC-2 TEM-1A*, OXA-9*, _SHV-182* >32 (R) >32 (R) >32 (R)

ARU871 K. pneumoniae KPC-2 CTX-M-15, TEM-1A, CMY-2, OXA-9*, OXA-10, SHV-13*, LEN12* >32 (R) >32 (R) >32 (R)

ARU919 K. pneumoniae KPC-2 CTX-M-65, _{TEM-1B*, SHV-12} >32 (R) >32 (R) >32 (R) ARU920 K. pneumoniae KPC-2 TEM-1A*, OXA-9*, _SHV-12 >32 (R) >32 (R) >32 (R) ARU725 K. pneumoniae NDM-1 CTX-M-15, TEM-1B, _SHV-12 >32 (R) >32 (R) >32 (R) ARU726 K. pneumoniae NDM-1 CTX-M-15, SHV-67* >32 (R) 32 (R) 16 (R) ARU884 K. pneumoniae NDM-1 CTX-M-15, TEM-1B, _{OXA-1, OKP-A-8} 32 (R) 8 (R) 8 (I) ARU923 K. pneumoniae NDM-1 CTX-M-15*, _{OXA-1*, SHV-94*} >32 (R) 16 (R) 8 (I) ARU928 K. pneumoniae NDM-5 CTX-M-15, TEM-1B, _{OXA-1*, SHV-13*} >32 (R) >32 (R) 32 (R) ARU724 K. pneumoniae NDM-1*, OXA-48 CTX-M-15, TEM-1B, _{OXA-1, SHV-67*} >32 (R) 32 (R) 16 (R) ARU879 K. pneumoniae NDM-1, OXA-48 CTX-M-15, TEM-1B, _{OXA-1, SHV-28*} >32 (R) >32 (R) >32 (R) ARU882 K. pneumoniae NDM-1, OXA-48 CTX-M-15, SHV-28* >32 (R) >32 (R) >32 (R) ARU734 K. pneumoniae OXA-48 CMY-4, SHV-172* >32 (R) 32 (R) >32 (R)

ARU735 K. pneumoniae OXA-48 CMY-4, SHV-81* 4 (R) 1 (S) 1 (S)

(18)

3.3 oCelloScope and spot test results

The main purpose of this study was to screen for synergistic effects among combinations of the carbapenems ertapenem, doripenem and meropenem. Clinically relevant concentrations of antibiotics were used, with ranges around the MIC values. Some strains (ARU887, ARU888, ARU889, ARU890, ARU891, ARU894 and ARU903) were tested with ranges of meropenem above the MIC values (section 2.2) in the ertapenem and meropenem combination, because the idea with using lower ranges of antibiotics came up after the first experiments. However, a time-lapse microscope (oCelloScope) was used to screen for synergies. In the oCelloScope, a camera was used to take images of the samples, and GKA algorithms (section 3.2) were used to illustrate bacterial growth. A 96-well microplate was loaded with antibiotics and bacteria, and the plate was incubated in the oCelloScope. The oCelloScope screening was followed by a spot test, which essentially means dilution and plating of the oCelloScope samples, in order to increase the resolution and verify the results. In all screenings, the P. aeruginosa reference strain was used instead of the E. coli or K. pneumoniae, since the MIC values matched the used concentration ranges in the screening better. Furthermore, a summary of all classification results (i.e. synergy, antagonism and indifference) from the oCelloScope screening and the spot test are shown in Table 3, as well as the strains’ carbapenemase(-s) and if any combination showed a bactericidal effect at one or more concentration. In most cases (25/30 strains), the effect of the combination therapy was indifferent compared to monotherapy. Interestingly, one or more antibiotic combinations showed a synergistic effect against 12/30 strains according to the spot test (six E. coli and six K. pneumoniae strains). If analyzing the oCelloScope results, synergy was observed against 11/30 strains. However, antagonism was seen against 8/30 strains according to the spot test, and against no strain according to the oCelloScope data. Generally, bactericidal effect was seen (in at least one well) against the strains also associated with synergistic effects. In some cases, bactericidal effect was observed even though the effect of the antibiotic combination was indifferent. In case of antagonism, bactericidal effect was not seen in the well with antagonism but at higher concentrations of antibiotics. One exception was seen, where the bactericidal effect was observed in the antagonistic well (ARU888 with ETP+DOR).

As mentioned in section 2.7, the custom script in Python generated images and graphs from the oCelloScope screening. An example of images and graphs is shown in Figure 7, where ARU923 was tested with the combination of doripenem and meropenem. As seen in Figure 7A, all wells had defined growth (i.e. reached the cut-off values for both BCA at 24 h and SESAmax) except

(19)

Table 3

Summary table of all spot test and oCelloScope screening classification results (in parenthesis) and what combinations of ertapenem (ETP), doripenem (DOR) and meropenem (MEM) that resulted in bactericidal effect in at least one well. The classifications were defined as synergy (S), antagonism (A) or indifference (I).

Classification of effect

Strain Carbapenemase _gene(-s) Bactericidal _{(What combination)} ETP+MEM ETP+DOR DOR+MEM

E.

co

li

ARU887 KPC-2 Yes (all) I (I) I (I) I (I)

ARU888 KPC-2 Yes (ETP+MEM and _ETP+DOR) I (I) A (I) S (S)

ARU894 KPC-2 Yes (ETP+MEM) I (I) I (I) I (I)

ARU711 NDM-1 No I (I) I (I) I (I)

ARU714 NDM-1 No A (I) I (I) I (I)

ARU709 NDM-5* No I (I) I (I) I (I)

ARU889 OXA-48 Yes (all) I (I) S (S) I (I)

ARU890 OXA-48 Yes (ETP+MEM and _DOR+MEM) S (I) S (S) A (I)

ARU891 OXA-48 Yes (ETP+MEM and _DOR+MEM) S (S) S (S) S (S)

ARU896 OXA-48 Yes (all) S (S) S (S) S (S)

ARU903 OXA-48 Yes (all) S (S) S (I) S (I)

K.

pne

um

oni

ae

ARU737 KPC-2 No I (I) I (I) I (I)

ARU920 KPC-2 No I (I) A (I) I (I)

ARU726 NDM-1 Yes (ETP+DOR) I (I) I (I) I (I)

ARU884 NDM-1 Yes (ETP+MEM and _DOR+MEM) I (I) I (I) I (I)

ARU923 NDM-1 No S (S) I (I) S (S)

ARU928 NDM-5 Yes (ETP+MEM) S (S) I (I) S (S)

ARU724 NDM-1* + OXA-48 Yes (DOR+MEM) A (I) I (I) S (S)

ARU879 NDM-1 + OXA-48 No A (I) I (I) I (I)

ARU882 NDM-1 + OXA-48 No I (I) I (I) A (I)

ARU734 OXA-48 Yes (all) I (I) S (I) I (I)

ARU735 OXA-48 Yes (all) S (S) I (I) S (S)

ARU736 OXA-48 Yes (all) A (I) S (I) S (S)

(20)

Figure 7

oCelloScope images and graphs of ARU923 with the doripenem (DOR) and meropenem (MEM) combination.

A) Images at 24 h. Synergistic activity can be seen with DOR 32 mg/L and MEM 64 mg/L – the highest

concentrations of antibiotics. Wells with growth, according to BCA and SESAmax cut-off values, are marked

with red squares. The green square indicates that neither BCA nor SESAmax were fulfilled (undefined growth).

B) The change of BCA values over time in the corresponding wells as seen in A. The increase in the BCA value

of the well with highest concentrations of antibiotics can be due to condensation or evaporation of MH-II broth.

According to the spot test, one or more antibiotic combinations showed synergy against 12 strains. These 12 strains are presented in Table 4, where the lowest concentrations at which synergy occurred are reported. Most synergies could be observed at relatively low concen-trations of antibiotics, while other strains required the highest concenconcen-trations to be affected (marked with orange in Table 4).

Table 4

Presentation of the lowest concentrations at which a synergistic effect was seen in the spot test (strains against which only indifferent effect was seen are excluded). If synergy was not obtained, the classification is written instead. Text in orange indicates that only the highest concentrations tested resulted in synergy.

Antibiotic combination

Strain ETP+MEM _(mg/L) ETP+DOR _(mg/L) DOR+MEM _(mg/L)

E.

co

li

ARU888 Indifference Antagonism DOR 8 + MEM 16

ARU889 Indifference ETP 4 + DOR 8 Indifference

ARU890 ETP 16 + MEM 2 ETP 16 + DOR 8 Antagonism

ARU891 ETP 16 + MEM 2 ETP 4 + DOR 8 DOR 0.125 + MEM 16

ARU896 ETP 0.5 + MEM 2 ETP 16 + DOR 0.125 DOR 1 + MEM 2

ARU903 ETP 4 + MEM 2 ETP 16 + DOR 0.125 DOR 8 + MEM 0.25

K

. pne

um

oni

ae

ARU724 Indifference Indifference DOR 1 + MEM 64

ARU734 Indifference ETP 16 + DOR 1 Indifference

ARU735 ETP 4 + MEM 2 Indifference DOR 1 + MEM 2

ARU736 Antagonism ETP 0.5 + DOR 32 DOR 8 + MEM 16

ARU923 ETP 16 + MEM 64 Indifference DOR 32 + MEM 64

ARU928 ETP 0.5 + MEM 64 Indifference DOR 32 + MEM 64

(21)

The aim was to screen for synergistic effects among carbapenem combinations using mainly the oCelloScope and spot tests. The effect from the combinations was hence defined as synergistic, indifferent or antagonistic. In Table 5, Table 6 and Table 7, the effects from the different combinations are presented in terms of BCA at 24 h and SESAmax for the oCelloScope

results and difference in log10 CFU/mL values for the spot test results. The data represent the

well where synergy was observed. If no synergistic effect could be seen, the well with highest concentrations of antibiotics (with defined growth) was picked instead. Moreover, the efficacy classification according to the oCelloScope data is also shown in Tables 5, 6 and 7. Full data can be seen in the Supplementary material, Tables S1, S2 and S3. In total, synergy against seven strains was seen with the combination of ertapenem and meropenem (four E. coli and three K. pneumoniae), whereof six could be detected with the oCelloScope screening (Table 5). The synergistic effect against ARU890 could be detected with the spot test, but not with the oCelloScope. Additionally, the ertapenem and meropenem combination showed antagonistic effect against ARU736, but only according to the spot test. Moreover, Table 6 displays the oCelloScope and spot test results from the combination of ertapenem and doripenem. A total of seven synergies were observed with this combination (five E. coli and two K. pneumoniae), but only four could be detected with the oCelloScope screening. Moreover, in three cases, the effect of the combination was classified as indifferent with the oCelloScope screening, while the spot test showed synergistic effect (ARU734, ARU736 and ARU903). The effect against ARU888 was antagonistic with the ertapenem and doripenem combination, but only according to the spot test. For the combination of doripenem and meropenem (Table 7), a total of nine synergies were identified with the spot test (four E. coli and five K. pneumoniae), whereof eight could be detected also with the oCelloScope. The combination was synergistic against ARU903 in the spot test, but not with the oCelloScope screening. According to the spot test, antagonism was observed against one strain (ARU890).

Table 5

oCelloScope and spot test results of the antibiotic combination ertapenem (ETP) and meropenem (MEM) against six E. coli and six K. pneumoniae strains. From the oCelloScope screening, BCA values at 24 h and SESAmax

are presented. Values that reached the cut-off value are marked with gray background (³ 8.00 for BCA and ³ 5.80 for SESAmax). Growth was defined if both BCA and SESAmax were fulfilled. Synergistic effects (S)

(≥ 2 log10 decrease in CFU/mL (stated as a negative value) in the spot test, or both BCA and SESAmax values

below cut-off for antibiotic combination and above cut-off for monotherapies) are marked with green background. Antagonistic effects (≥ 2 log10 increase in CFU/mL or both BCA and SESAmax values above cut-off

for antibiotic combination and below cut-off for any monotherapy at the same concentration) are marked with red background. I = Indifferent effect (< 2 log10 decrease or increase in CFU/mL at 24 h or both BCA and

SESAmax above cut-off for the combination as well as the monotherapies at the same concentrations).

ETP monotherapy MEM monotherapy _combinationETP+MEM

(22)

Table 5

Continued

ETP monotherapy MEM monotherapy _combinationETP+MEM _Spot test Δlog10 CFU/ mL Classi-fication (BCA/ SESA) ARU BCA at 24 h SESAmax Spot test log10 CFU/ mL BCA at 24 h SESAmax Spot test log10 CFU/ mL BCA at 24 h SESAmax Spot test log10 CFU/ mL K . pne um oni ae 724 8,95 6,28 8 8,95 6,24 8 8,98 6,22 7 -1 I 734 8,96 6,31 8 8,94 6,32 8 8,95 6,27 7 -1 I 735 8,97 6,19 8 8,92 6,23 8 6,43 4,92 < 2 £ -7 S 736 8,92 6,33 8 8,87 6,33 7 8,97 6,21 9 2 I 923 8,94 6,31 7 9,00 6,21 8 8,77 4,70 < 2 £ -6 S 928 9,00 6,30 6 9,00 6,27 7 8,83 5,06 < 2 £ -5 S Table 6

oCelloScope and spot test results of the antibiotic combination ertapenem (ETP) and doripenem (DOR) against six E. coli and six K. pneumoniae strains. From the oCelloScope screening, BCA values at 24 h and SESAmax are

presented. Values that reached the cut-off value are marked with gray background (³ 8.00 for BCA and ³ 5.80 for SESAmax). Growth was defined if both BCA and SESAmax were fulfilled. Synergistic effects (S)

ETP monotherapy DOR monotherapy _combinationETP+DOR

(23)

Table 7

oCelloScope and spot test results of the antibiotic combination doripenem (DOR) and meropenem (MEM) against six E. coli and six K. pneumoniae strains. From the oCelloScope screening, BCA values at 24 h and SESAmax are presented. Values that reached the cut-off value are marked with gray background (³ 8.00 for BCA

and ³ 5.80 for SESAmax). Growth was defined if both BCA and SESAmax were fulfilled. Synergistic effects (S)

DOR monotherapy MEM monotherapy DOR+MEM _combination _Spot test Δlog10 CFU/ mL Classi-fication (BCA/ SESA) ARU BCA at 24 h SESA max Spot test log10 CFU/ mL BCA at 24 h SESA max Spot test log10 CFU/ mL BCA at 24 h SESA max Spot test log10 CFU/ mL E. co li 888 9,01 6,17 8 9,02 6,16 9 7,25 5,32 6 -2 S 889 8,98 6,15 8 9,04 6,24 9 8,87 6,10 8 0 I 890 8,95 6,24 7 8,97 6,28 8 8,95 6,20 9 2 I 891 9,04 6,19 9 8,94 6,12 9 8,82 4,77 < 2 £ -8 S 896 8,98 6,30 8 8,96 6,23 8 8,79 5,15 6 -2 S 903 9,02 6,30 8 9,03 6,24 8 9,02 6,15 < 2 £ -7 I K . pne um oni ae 724 9,02 6,27 8 9,00 6,16 7 7,96 4,92 < 2 £ -6 S 734 9,02 6,34 9 8,99 6,36 8 8,99 6,23 9 1 I 735 8,96 6,38 8 8,99 6,19 8 8,80 4,89 < 2 £ -7 S 736 8,97 6,27 8 8,93 6,36 8 7,94 4,42 < 2 £ -7 S 923 8,94 6,26 8 8,98 6,21 8 8,08 4,75 < 2 £ -7 S 928 8,97 6,30 7 8,99 6,23 6 8,72 4,48 < 2 £ -5 S Table 8

Summary of the effect according to the spot test from all antibiotic combination therapies with K. pneumoniae (E. coli strains in parenthesis). The numbers refer to the number of strains that synergy, indifference or antagonism was seen against, with all antibiotic combinations. Three carbapenem combinations were screened against a total of 30 strains. No E. coli strain carried both the NDM and OXA-48 enzymes.

All three antibiotic combinations

Carbapenemase Synergy Indifference Antagonism

NDM 4 (0) 11 (17) 0 (1)

KPC 0 (1) 14 (7) 1 (1)

NDM + OXA-48 1 (0) 5 (0) 3 (0)

OXA-48 5 (12) 3 (2) 1 (1)

Total: 10 (13) 33 (26) 5 (3)

(24)

against strains carrying the OXA-48 carbapenemase gene (5/10 cases with K. pneumoniae strains and 12/13 cases with E. coli strains). However, synergies could also be seen against strains carrying NDM – but only regarding K. pneumoniae (Table 3; ARU724, ARU923 and ARU928). Synergy was observed against one strain carrying KPC (Table 3; E. coli ARU888 with DOR+MEM). Moreover, antagonism was seen against a total of five K. pneumoniae strains, carrying either KPC, OXA-48 or both NDM and OXA-48. When it comes to E. coli, the antibiotic combinations showed antagonism against three strains, which carried either NDM, KPC or OXA-48.

3.3 Carry-over effects

When transferring the samples from the wells not only bacteria, but also the antibiotics are transferred to the agar plate, there is thus a risk that the antibiotics can prevent or reduce growth on the plate. This is referred to as a carry-over effect.36_{In case of this project, the high}

concen-trations of carbapenems could potentially have an effect on the bacteria when suspensions were plated onto agar plates. The lasting carbapenem effect could hinder the upturn of viable colo-nies, which in turn could lead to an overestimation of the efficacy of carbapenem combination therapy.37,38_{To ensure that the lack of bacterial growth in the spots was not due to a carry-over}

effect of antibiotics, four strains (two E. coli and two K. pneumoniae) that showed suspected carry-over in the spot test (i.e. mainly growth around the edges of the spot) were further analyzed by a carry-over test. This test is performed by comparing growth in spot test and growth on separate agar plates with regular plating using beads. Carry-over effect would be noticed as deviations in log10 CFU/mL of the spot test and glass bead plating or inhibition zones

in the dense growth of bacteria plated with beads. oCelloScope samples were diluted up to 100-fold, since any higher dilution would be well below the MIC values. However, none of the strains showed deviations > 1 log10 CFU/mL between the two methods (data not shown), which

was considered to be an acceptable variation. No inhibition zones could be identified.

3.4 Time-kill results

Time-kill assays were performed in order to further evaluate the oCelloScope results, by taking samples at specific time points from cultures containing different concentrations of antibiotics. Synergy, antagonism and indifference were defined in the same way as for the spot test (section 3.2). The spot test demonstrated that one or more antibiotic combinations showed synergy against three NDM-producing strains (ARU724, ARU923 and ARU928; ARU724 also OXA-producing). However, when further analyzed in time-kill, all carbapenem combinations showed an indifferent effect, meaning that the two methods did not agree with each other. Moreover, the growth curve for ARU724 is illustrated in Figure 8A. The growth was inhibited to essen-tially the same level by all the antibiotics and combinations in the beginning of the assay (at the 2 h sampling point) compared to the growth control. However, the bacteria re-grew over time both in single antibiotic and combination tubes. At 24 h, the re-growth had reached the same level as the control tube. The pattern with re-growth was present also for ARU923 (Figure 8B). Regarding ARU928 (Figure 8C), the effect from two antibiotic combinations (ETP+MEM and DOR+MEM) and one single antibiotic (MEM) inhibited re-growth of the bacteria for a longer time, and they did not re-grow to the same extent as ARU724 and ARU923. Furthermore, the greatest decrease in log10 CFU/mL was seen with the DOR+MEM combination, which showed

(25)

Figure 8

Time-kill curves on average log10 CFU/mL values from duplicates, for A) ARU724, B) ARU923 and

C) ARU928. For all tested strains, a reduction in the number of bacteria was seen at the 2 h sampling point.

However, re-growth was seen in the tubes containing antibiotics, leading to no synergistic effects from any antibiotic combinations. Concentrations given in the labels to the right are in the unit mg/L.

(26)

4 Discussion

In this study, combinations of the carbapenems ertapenem, doripenem and meropenem were screened for synergistic effects against 30 strains of E. coli and K. pneumoniae. The main purpose with this study was to add more knowledge and data regarding double carbapenem therapy against multidrug resistant bacteria. The methods used were mainly time-lapse micro-scopy (oCelloScope) followed by so-called spot tests. Generally, the oCelloScope results agreed with those of the spot test (Table 3). On the other hand, no antagonistic effects could be detected in the oCelloScope screening, while antagonism was present against 8/30 strains according to the spot test. Synergistic effects were seen in a total of 23/90 cases according to the spot test (90 because of three antibiotic combination screened against a total of 30 strains). In five of these 23 cases (ARU890 with ETP+MEM; ARU734, ARU736 and ARU903 with ETP+DOR; ARU903 with DOR+MEM), the oCelloScope screening was unable to detect the synergies. In four of these five cases, the growth was undefined in both the combination well and single antibiotic well according to the oCelloScope data (Table 3 and Supplementary material, Table S1, Table S2 and Table S3). When looking at the oCelloScope images (data not shown), no growth was visualized. The images do not show the whole well though, which means that growth cannot be precluded. Moreover, the fifth strain had defined growth in both the combination well and the single antibiotic well, which also could be seen on the oCello-Scope images. Therefore, by performing a spot test, the amount of growth could be obtained because of a higher resolution. The detection level for the oCelloScope was 5 log10 CFU/mL,

while it was 2 log10 CFU/mL for the spot test. In other words, if a well had growth below

5 log10 CFU/mL, the growth was undefined according to the oCelloScope data, while the growth

would be identifiable in the spot test. Furthermore, the appearance of growth in the two methods could also be affected by the different media (MH-II broth in oCelloScope and MH-II agar plates in spot test).

As mentioned above, antagonism was observed against 8/30 strains with the spot test. However, the antagonistic effect seemed to pop up without any specific pattern (Supplementary material, Table S4, Table S5 and Table S6). In most cases (5/8 strains), the reason for the ³ 2 log10

increase in CFU/mL more likely depended on a slight deviation of lower growth in the single antibiotic well (1-2 log10 CFU/mL lower than in the surrounding wells). An example of this

was seen with ARU724 with the ertapenem and meropenem combination, where the well with only ETP 4 mg/L reached 7 log10 CFU/mL, while both ETP 0.5 mg/L and ETP 16 mg/L reached

8 log10 CFU/mL (Supplementary material, Table S1). If the growth had reached 8 log10

CFU/mL also in the well with only ETP 4 mg/L, the combination would not be defined to have an antagonistic effect against ARU724 but an indifferent effect. These findings suggest that further analysis is required to draw the conclusion that there was an actual antagonistic effect between the antibiotics. Perhaps the spot test is too crude and one should consider increasing the limits to 3 log10 CFU/mL difference.

(27)

always identified against the same ARU numbers (Table 3). According to the intended mechanism of action seen in Figure 2, combinations that included ertapenem would be most efficient, although this was obviously not the case. Differences in efficacy of the different combinations can be related to other resistance mechanisms such as porin deficiencies or presence and enhanced production of efflux pumps. Furthermore, most strains (25/30 strains) carried at least one b-lactamase gene besides the carbapenemase gene. b-lactamases other than carbapenemases often lead to resistance to different kinds of penicillins and cephalosporins, but not to carbapenems, and they are likely to co-occur in the same bacterium.37,39_Some

b-lactamase genes (e.g. OXA-1 or SHV) are often carried on plasmids and other b-lactamases are almost always carried chromosomally (e.g. LEN or OKP).40,41_{The carriage of other}

b-lacta-mases could potentially have an impact on how well the antibiotic combinations worked, if the b-lactamases e.g. bind to carbapenems to some extent. Furthermore, if ertapenem, doripenem and meropenem bind to different PBPs, this would also potentially affect the efficacy of a combination. In fact, ertapenem and meropenem bind with high affinity to PBP-2, followed by PBP-3, and doripenem binds most strongly to PBP-2 in E. coli. Additionally, ertapenem and meropenem can also bind to PBP-1a and PBP-1b with strong affinity.42_{In other words, the}

variabilities of PBP-bindings can influence how effective the combination is in terms of synergistic effect.

The concentrations of antibiotics needed for synergistic effects varied among screenings (Table 4). In some cases, the highest tested concentrations were required in order to eliminate the bacteria, but in most cases, synergy was seen at lower concentrations (although all concentrations used were clinically relevant). This relates to the MIC values of the bacteria, since bacteria with lower MIC values should be easier to wipe out with lower concentrations of antibiotics. Actually, all E. coli strains that synergy was found against in the spot test (a total of six strains) were susceptible both to doripenem and meropenem. Interestingly, synergy was observed in 4/6 cases with the doripenem and meropenem combination (Table 4) – mainly against strains carrying only OXA-48. A possible cause for this pattern is the lower activity of OXA-48, leading to lower resistance levels. For the two other E. coli strains, indifference (ARU889) and antagonism (ARU890) were noticed. ARU889 was able to grow in wells with a doripenem concentration of 8 mg/L (Supplementary material, Table S3), even though the dori-penem MIC value was 1 mg/L (Table 2). However, synergies could also be observed at lower antibiotic concentrations even though the strain had MIC values of >32 mg/L. From a clinical perspective, combinations that show synergy at low antibiotic concentrations would be pre-ferred.

Commonly found, growth was observed in the oCelloScope also in wells with antibiotic concentrations above the MIC values (Table 2 and Supplementary material Table S1, Table S2 and Table S3). Theoretically, a strain should be eliminated if treated with antibiotics above the MIC values, like indicated above. However, this was thus not always the case. An explanation for this phenomenon is stated as the inoculum effect (IE), which is described as an increase of the MIC value due to a higher inoculum rather than actual resistance.43_{Another possible cause}

for the higher MIC value can be the longer incubation time of the oCelloScope screening compared to the BMD.

(28)

the single antibiotic therapy also inhibited growth besides the combination (appeared as an indifferent effect). An early effect from an antibiotic therapy could potentially be of great importance in clinical regimens, since a treatment that eliminates the bacteria fast is preferred. A reduced number of bacteria enables for the immune system to clear the infection together with further doses of antibiotics.

The agreement between the time-kill assay and the oCelloScope and spot test methods was poor for the selected antibiotic concentrations. Of the three strains analyzed in the time-kill experi-ments, no synergistic effects were observed because of re-growth. The strains were chosen based on that synergy had been seen against them, and that they carried the high-activity carbapenemase NDM. In the oCelloScope, BCA at 24 h and SESAmax were used to illustrate

bacterial growth. High BCA values can be due to filamentation of bacteria, condensation on the plastic film covering the plate or evaporation of liquid in the well cause a darker image. Hence, growth was only defined if both BCA at 24 h and SESAmax were above the chosen cut-off value.

Furthermore, re-growth has previously also been reported by Drusano (2004) in time-kill experiments, and reasons for this phenomenon might be a larger working volume in the time-kill assay compared to the oCelloScope. oCelloScope experiments were performed using a final volume of 200 µL, which was about 14-fold lower than the time-kill volumes. Larger starting volumes entail higher numbers of bacteria, meaning that the total number of carbapenemases also increases. The risk for pre-existing subpopulations of resistant bacteria increases with higher starting volumes, as well as the risk for emergence of resistance mutations during experiments.32_{Also, the bacteria had access to more oxygen due to the shaking incubation,}

which further promotes bacterial growth.

As a screening method for detecting antimicrobial activity in high-throughput screenings involving e.g. antibiotic combinations, the oCelloScope has the potential to assess synergistic effects. In order to see how the oCelloScope method generally compares to the time-kill method (golden standard), more strains have to be analyzed with the time-kill assay. On the other hand, the potentiality of using the oCelloScope as a high-throughput screening method is not affected by the poor matching between oCelloScope and time-kill results. This is because the oCello-Scope results cannot be compared directly to the time-kill results, since e.g. the detection level (5 log10 CFU/mL for oCelloScope and 1 log10 CFU/mL for time-kill) differs between the

methods. However, some advantages with the oCelloScope method are that it is an automated method and hence less time-consuming and labor-intensive than the time-kill assay. Many samples can be tested simultaneously, as the microplate has 96 wells. Since images are taken during the run, the oCelloScope can also be used in order to analyze morphological changes. Due to the rather high detection level, the oCelloScope was more prone to miss synergies compared to the spot test. Another disadvantage is that both live and dead cells, condensation and evaporation can affect how well bacteria are identified.

Overall, the data in this project supports the superiority of carbapenem combination therapy over monotherapy against a subset of strains – especially against strains carrying OXA-48. However, some of the highly resistant strains were also eliminated (mainly K. pneumoniae), which can potentially be of great importance for treatment of individual patients. An important aspect that needs to be studied further is if the observed decrease in log10 CFU/mL was due to

(29)

method. Other future studies involve e.g. determination of antibiotic concentrations at the end of the oCelloScope screening and the time-kill experiments, and studying a larger number of strains in the time-kill assay. Other possible resistance mechanisms should also be examined. Even further in the future, the carbapenem combination therapy could be tested against constructed strains with isogenic backgrounds, only differing in the resistance mechanisms, to explore mechanisms of resistance and genotype/phenotype associations.

5 Ethical approval

Ethical approval was not required for this project.

6 Conclusion

Synergistic effects of double carbapenem therapy were seen against 12/30 strains (six E. coli and six K. pneumoniae). Most synergies were found against strains which had lower MIC values, carrying only the OXA-48 carbapenemase gene. However, synergies were also found against highly resistant strains carrying the NDM or KPC gene. Furthermore, the oCelloScope has the potential of being used in high-throughput screenings, even though more studies have to be performed and compared with the golden standard (time-kill). Regarding carbapenem combination therapy, more studies are needed before applying it in clinical regimens.

Screening for double carbapenem therapies effective against multidrug resistant Enterobacteriaceae